Picomole-scale characterization of protein stability
and function by quantitative cysteine reactivity
Daniel G. Isom,1, Eyal Vardy,1, Terrence G. Oas, and Homme W. Hellinga2
Department of Biochemistry, Duke University, DUMC Box 3711, Durham, NC 27710
Edited by David Baker, University of Washington, Seattle, WA, and approved January 15, 2010 (received for review September 11, 2009)
The Gibbs free energy difference between native and unfolded
states (“stability”) is one of the fundamental characteristics of a
protein. By exploiting the thermodynamic linkage between ligand
binding and stability, interactions of a protein with small mole-
cules, nucleic acids, or other proteins can be detected and quanti-
fied. Determination of protein stability can therefore provide a
universal monitor of biochemical function. Yet, the use of stability
measurements as a functional probe is underutilized, because such
experiments traditionally require large amounts of protein and
special instrumentation. Here we present the quantitative cysteine
reactivity (QCR) technique to determine protein stabilities rapidly
and accuratelyusing only picomole quantities of material and read-
ily accessible laboratory equipment. We demonstrate that QCR-
derived stabilities can be used to measure ligand binding over a
wide range of ligand concentrations and affinities. We anticipate
that this technique willhave broad applications in high-throughput
protein engineering experiments and functional genomics.
conformational stability ∣ thermal stability ∣ ligand-binding affinity ∣
linkage analysis ∣ thiol protection
inhibitors, or activators; receptors with ligands; protein-protein
networks; protein-DNA; and protein-RNA). Such functional in-
teractions all affect protein stability by virtue of a thermodynamic
linkage relationship between the Gibbs free energy of folding
(ΔGU) and binding free energy (ΔGbind) of a ligand (L) to the
native (N) or denatured (D) states (1–3)
L þ D
Macromolecular stability is therefore one of the most fundamen-
tal thermodynamic measures in biochemistry by quantitatively re-
porting on structure-function relationships to provide a universal
monitor for biochemical function.
There are two distinct approaches for determining protein sta-
bility (4). The first measures the free energy of protein (un)fold-
ing under equilibrium conditions by assessing the fraction of the
native state using spectroscopy, hydrodynamic observations, func-
tional assays, or calorimetry. The second exploits the relationship
between protein dynamics and stability by monitoring the differ-
ential reactivity of internal chemical groups in native and un-
folded states. This second approach measures conformational
free energies, which under appropriate conditions corresponds
to global protein stability. Amide proton exchange is used most
commonly to monitor such differential reactivity (5–9), but its
widespread use to assess biological function typically is limited
by the need for specialized instrumentation and relatively large
amounts of protein. Recently, cysteine reactivity (10–14) and pro-
teolysis (15) have emerged as alternative means to determine
rates of protein (un)folding and estimate protein stabilities. Here
we present a method, quantitative cysteine reactivity (QCR), in
iomolecular function is most often the consequence of
interactions between molecules (enzymes with substrates,
N þ L
which protein stability is determined by monitoring the reactivity
of cysteine residues buried in the hydrophobic core of proteins.
This approach has the advantage over more traditional methods
for measuring protein stability in that it requires only picomoles
(nanograms) of protein, uses simple instrumentation accessible
to any lab, can be reasonably high throughput, and can provide
site-specific thermodynamic information. QCR can be used to
determine apparent protein stabilities rapidly and accurately,
construct Gibbs-Helmholtz stability profiles, measure ligand
binding over a large range of ligand concentrations and affinities,
and infer enzymatic activity without the need for developing a
kinetic assay. Here we demonstrate these capabilities by charac-
terizing three model proteins, Staphylococcal nuclease (SN),
Escherichia coli ribose-binding protein (ecRBP) and E. coli
maltose-binding protein (ecMBP), mutated to contain single,
buried cysteine residues.
Results and Discussion
Measuring Conformational Free Energies by QCR. The QCR experi-
ment is designed to determine the Gibbs free energy of global
protein unfolding by measuring the reactivity of wild-type or mu-
tant cysteines buried in a hydrophobic core. Here we demonstrate
the QCR approach in three model proteins (SN, ecRBP, and
ecMBP) mutated to contain cysteine probes located in internal
tact directly bound metals, inhibitors, or ligands (see Fig. S1).
These mutant proteins were produced in 200 μL batches by
cell-free coupled transcription and translation in E. coli extract.
Following affinity purification, which typically yields 0.4 to 1 μg
of protein, the reactivity of the buried cysteines was determined
tein species were separated by gel electrophoresis and quantified
by densitometry (Fig. 1). This approach requires approximately
0.5 picomoles (∼10 nanograms) of protein per time point and
can useavarietyofthiol-reactive reagentsthatalterelectrophore-
tic mobility or are highly fluorescent.
The QCR method exploits the conformational fluctuations of
a protein to measure conformational free energy. Cysteines that
are protected in the folded ensemble can be modified by
thiol-reactive probe only by complete exposure to bulk solvent
by transient unfolding reactions, as described by a two-step reac-
tion scheme (5, 13)
Unfolding free energies (ΔGU) can be determined under EX2
conditions (where kclose≫ kint) by measuring klabel, the rate
D.G.I., E.V., and T.G.O. analyzed data; and D.G.I., T.G.O., and H.H. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1D.G.I. and E.V. contributed equally to this work.
2To whom correspondence should be addressed. E-mail: firstname.lastname@example.org.
This article contains supporting information online at www.pnas.org/cgi/content/full/
4908–4913 ∣ PNAS ∣ March 16, 2010 ∣ vol. 107 ∣ no. 11 www.pnas.org/cgi/doi/10.1073/pnas.0910421107
constant for labeling a protected cysteine at a specified concen-
tration of thiol probe ½P? (5)
ðkopenþ kcloseþ kintÞ
ð1 þ eΔGU∕RTÞ;
where ΔGUis related to the closing (kclose) and opening (kopen)
reaction as ΔGU¼ RT lnkclose∕kopen, and kintis the product of ½P?
and the biomolecular rate constant for the reaction of an unpro-
tected cysteine (kint¼ k½P?). Values for kintcan be obtained from
the reactivity of unprotected cysteine residues in model com-
pounds or unfolded proteins for the accurate determination of
ΔGU(see Methods). Rearrangement of Eq. 3 yields conforma-
tional free energy
ΔGU¼ RT ln
A buried cysteine can be labeled as a result of local, subglobal,
or global unfolding transitions. The predominant mechanism of
cysteine modification can be converted from local or partial un-
folding to global unfolding by setting up conditions under which
global stability is diminished (e.g. by addition of denaturation or
by increasing temperature) (16, 17). We refer to the range of con-
ditions under which access to global unfolding predominates as
To ensure that the buried cysteines report global free energies
(i.e. ΔGU), QCR experiments are always performed within a
GUWO. We have chosen to use temperature to access the
GUWO and measure global unfolding free energies as a function
of temperature (ΔGUðTÞ), described by the Gibbs-Helmholtz
where ΔHmis the enthalpy of unfolding, ΔCpthe change in heat
capacity of unfolding, and Tmthe midpoint of thermal denatura-
tion, readily determined by QCR as the temperature at which
kint¼ 2klabel. The temperature range over which observations
can be made is determined by the limits where differences be-
tween klabeland kintexceed experimental error, EX2 conditions
prevail, and the GUWO is present (see Fig. 2). This range com-
prises a small portion of a Gibbs-Helmholtz curve. Consequently,
values for ΔHmand ΔCpderived from a fit of the temperature
dependence of ΔGUare usually underdetermined (19, 20), and
values for ΔCpmust be assigned a priori to derive reasonable
estimates for ΔHmand Tmfrom stabilities measured within
Usinga total of only
(∼2.5 picomoles or ∼50 nanograms per temperature point),
Gibbs-Helmholtz profiles were determined for two cysteine mu-
tants of SN and ecRBP (Fig. 3). Derived values for ΔHmand Tm
were relatively insensitive to ΔCpvalues within the range of
ΔGUðTÞ ¼ ΔHm
ðTm− TÞ þ T lnT
47.1 °C, pH 7.6 (left) (labeling times indicated for lanes 2–6). Streptavidin was used to alter the electrophoretic mobility of the labeled protein (streptavidin
bands indicated by a, b, and c). Unlabeled fractions were quantified by densitometry and fit with a single exponential to obtain reaction rates (right) at
different temperatures (54.6°C, blue; 51.7 °C, green; 48.9°C, red; 47.1 °C, orange; 45.2 °C, purple; 44.5°C, black); corresponding reaction rates are
2.6 × 10−3, 1.9 × 10−3, 9.3 × 10−4, 5.2 × 10−4, 2.6 × 10−4, and 2.0 × 10−4s−1, respectively. Error bars represent the estimated uncertainty of the integrated band
intensities (∼2%). (B) Labeling of SN variant L36C with IAM-biotin at 35.3 °C, pH 7.6 (left). The (un)labeled forms migrate differently in the gel, enabling
ratiometric quantification to obtain reaction rates (right) at different temperatures (38.3°C, blue; 35.3 °C, green; 32.3 °C, red; 29.3 °C, orange; 26.3 °C, purple;
23.3 °C, black); corresponding reaction rates are 9.2 × 10−4, 3.6 × 10−4, 1.2 × 10−4, 1.5 × 10−4, 7.2 × 10−5, and 3.1 × 10−5s−1, respectively. At 29.3 °C, 26.3 °C, and
23.3 °C, kintand klabelwere manipulated by increasing the concentration of IAM-biotin from 1 mM to 3.16 mM.
Representative QCR experiments for ecRBP and SN. (A) SDS-PAGE of time-dependent modification of ecRBP variant L62C with 1 mM IAM-biotin at
Isom et al. PNAS
March 16, 2010
2–5 kcalmol−1K−1, which is consistent with previous experimen-
tally determined values for proteins in general (21–23). All four
cysteine mutants are thermally destabilized: the apparent Tm
values of SN variants F34C (40 ? 1°C) and L36C (39 ? 1°C)
are ∼13.0°C below wild-type (53.0°C) (24); the apparent Tmva-
lues of ecRBP variants L62C (54 ? 1°C) and A188C (56 ? 1°C)
are ∼8°C below wild-type (62.6°C) (25). The extrapolated
72 ? 1
0.1 kcalmol−1and 2.6 ? 0.1 kcalmol−1, whereas the stability of
5.5 ? 0.1 kcalmol−1(26). Similarly, the ΔGo
mutants L62C and A188C, using ΔHmvalues of 81 ? 2 and
91 ? 4 kcalmol−1,
4.1 ? 0.1 kcalmol−1, whereas the stability of wild-type reported
by chemical denaturation is 5.9 ? 0.4 kcalmol−1(27). This de-
crease in stability caused by the introduction of cysteine is typical
for mutations in the hydrophobic core of these (26, 28) and other
Uat 20 °C for SN mutants F34C and L36C, using ΔHmvalues
71 ? 1 kcalmol−1, respectively, is
Uat 25°C for ecRBP
3.2 ? 0.1 kcalmol−1
Measuring Ligand Affinity by Linkage Analysis of Protein Stability.
The modulation of protein stability by binding of metals, ligands,
activators, inhibitors, substrates, nucleic acid, or other proteins
can be used to measure binding affinities within a GUWO.
For a protein with a single binding site, the free energy of ligand
binding is described by
ΔGbind¼ RT lnP ¼ RT ln
where P is the binding polynomial (1–3), ½L? the total ligand
concentration, and KDthe apparent dissociation constant of
the ligand. For proteins with multiple ligand-binding sites, P is
expanded (SI Text). By thermodynamic linkage (3, 30) any change
in ΔGUcaused predominantly by ligand binding (Eq. 1) is
ΔGbind¼ ΔΔLGU¼ ΔLGU− ΔapoGU;
where ΔLGUand ΔapoGUare the stability of the protein in the
presence or absence of ligand, respectively. Eq. 7 is used to obtain
apparent KDvalues from either the ligand dependence of ΔΔGU
(by curve fitting) or from a single measurement of ΔΔGU.
Both ecRBP and ecMBP have a single ligand-binding site
located within the interface between their N-terminal and
C-terminal domains. With ∼10 picomoles (∼200 nanograms) of
protein, binding affinities were determined by QCR experiments
using two independent cysteine reporters introduced into each
domain. ecRBP variants L62C (N-terminal domain) and
A188C (C-terminal domain) report ribose-binding affinities of
1.5 ? 0.2 μM (at 48.9°C) and 1.8 ? 0.1 μM (at 54.6°C), respec-
tively (Fig. 4A); ecMBP variants T157C (C-terminal domain) and
S263C (N-terminal domain) report maltose-binding affinities of
8.0 ? 0.2 μM and 11.8 ? 0.8 μM, respectively at 63.3°C (Fig. 4B).
Proteins that bind more than one ligand, either independently
or synergistically,have amorecomplex freeenergylandscape that
involves a number of different ligand-bound species (SI Text).
Here, we demonstrate how QCR can be used to characterize
unfolding free energies (ΔGU) can be determined by quantitative cysteine
reactivity. The first limits are set by the accuracy of the measurement of
the labeling rate constants: an upper limit occurs at a temperature
(maxTexp) and free energy (minΔGexp) at ∼10°C above Tm(red-dashed arrows)
where the difference of klabeland kintis within experimental error; a lower
limit occurs at a temperature (minTexp) and free energy (maxΔGexp) at
∼10–20°C below Tm(green-dashed arrows) where increased stability suffi-
ciently reduces klabel(Eqs. 3 and 4) such that it appears to be independent
of temperature within experimental error. The second limit is set in some
cases where the mechanism of cysteine protection (i.e. local or global unfold-
ing) is dependent on temperature. Such cases manifest themselves as a de-
viation of the observed temperature dependence of ΔGUfrom that expected
for global unfolding. It is well established that global unfolding conditions
prevail within ∼10–20°C of Tm(16, 17), which we refer to as the global un-
folding window of observation. The black line illustrates a case in which there
is no such switch (modeled by Eq. 5) and the GUWO extends over the entire
temperature range; the gray line represents switching between global and
local unfolding with a concomitant temperature limit for the GUWO (mod-
eled by Eq. 12 of ref. 6). The third limit is set at a point where EX1 conditions
prevail and kcloseno longer exceeds kint(not illustrated). This may occur as
stability is diminished (ΔGU< 1 kcal∕mol) or if the concentration of thiol
probe ½P? is too high. Loss of EX2 conditions is manifested as a loss of the
linear dependence of klabelon ½P? and can be remedied by reducing ½P?.
The overall temperature range at which observations can be made is the
intersection of all three of these conditions (black and gray bars).
Three factors determine the temperature range at which global
variants F34C (purple) and L36C (black), and (B) ecRBP variants L62C (black)
and A188C (purple). Solid lines indicate a fit to a Gibbs-Helmholtz profile
(Eq. 5) using a fixed ΔCpof 3 kcalmol−1K−1. Error bars represent the error
of three independent experiments at select temperatures.
Temperature dependence of ΔGUdetermined by QCR for (A) SN
www.pnas.org/cgi/doi/10.1073/pnas.0910421107Isom et al.
the binding of a calcium ion (Ca2þ) and a 5′-monophosphate
inhibitor (pdTp) to SN. By themselves, Ca2þand pdTp bind to
SN with affinities of ∼500 μM and ∼90 μM (31). Using
∼5 picomoles (∼100 nanograms) of protein, the KDvalues of
each binary complex was determined by QCR (Fig. 5A). The
increase in stability of SN/L36C in the presence of 1 mM Ca2þ
or 50 μM pdTp is 0.6 ? 0.1 and 0.7 ? 0.1 kcal∕mol, which cor-
23 ? 4 μM, respectively. Binding of Ca2þand pdTp is synergis-
tic, as each exhibits a greater affinity (∼2 μM) for SN in the
presence of the other (31). Consequently, a 2∶1 molar solution
ligand Ca2þ-pdTp. Using ∼10 picomoles (∼200 nanograms) of
protein, the affinity of Ca2þ-pdTp for SN variants F34C and
L36C was determined to be 4.8 ? 0.2 μM and 2.2 ? 0.1 μM,
respectively, at 35.3°C (Fig. 5B).
600 ? 200 μM and
Inferring Enzyme Activity by QCR. Binding of substrates and
products also affects enzyme stability in a detectable manner
(32). QCR therefore provides a means to infer enzymatic activity
within a GUWO using only picomole quantities of protein with-
out having to devise a reaction-specific kinetic assay. We demon-
strate this approach using SN, a 5′-phosphodiesterase that
hydrolizes single- and double-stranded DNA and RNA. It selec-
tively cleaves the phosphodiester bond between the phosphate
and 5′-hydroxyl, producing short 3′-derived oligonucleotides
(which do not bind to SN) and 5′-derived mononucleotides
(which bind to and inhibit SN) (33). In the absence of calcium,
L62C (purple) at 48.9 °C and A188C (black) at 54.6 °C, and for (B) ecMBP
variants T157C (black) and S263C (purple) at 63.3°C. The solid lines represent
the fit of Eq 7 to the data to obtain KDvalues. Error bars correspond to the
propagated uncertainty of two combined ΔGUmeasurements.
Ligand concentration dependence of ΔΔGUfor (A) ecRBP variants
for SN variant L36C in the absence (black) and presence of 1 mM Ca2þ(purple)
or 50 μM pdTp (orange) at 35.3 °C. Observed rate constants of 3.0 × 10−4,
1.2 × 10−4, and 9.2 × 10−5s−1, respectively, correspond to ΔΔGUvalues of
0.6 ? 0.2 and 0.7 ? 0.2 kcalmol−1. (B) Dependence of ΔΔGUon a 2∶1 molar
ratio of Ca2þand pdTp for SN variants F34C (black) and L36C (purple) fit with
Eq. 7 to obtain KDvalues. (C) QCR experiments at 35.3°C for SN variant L36C
in the absence of substrate (black) and 4.7 μM single-stranded DNA (green),
4.7 μM single-stranded DNA with 1 mM Ca2þ(blue), and 12 μM of a 2∶1 molar
ratio of Ca2þand pdTp (red). Observed rate constants of 3.0 × 10−4,
2.8 × 10−4, 7.5 × 10−5, and 5.9 × 10−5s−1, respectively, correspond to ΔΔGU
values of 0.1 ? 0.2, 0.9 ? 0.2, and 1.0 ? 0.2 kcalmol−1. The L36C mutant is
enzymatically active (inset; 1% agarose gel): 1.5 kb double-stranded DNA
fragment (lane 2) digested completely (lane 1) by incubation with 0.05 μM
SN/L36C at 20°C for 10 min in a buffer of 1 mM Ca2þ, 25 mM MOPS,
100 mM KCl, and pH 7.6. The kcat, kcat∕Km, and Kmof SN for canonical sub-
strate (double-stranded salmon sperm DNA) are ∼90 s−1, 2 × 106M−1s−1, and
∼50 μM, respectively, at pH 7 (38).
Effect of ligands and substrate on SN stability. (A) QCR experiments
Isom et al.PNAS
March 16, 2010
SN is inactive and the binding of substrate alone can be measured
by QCR. Addition of 4.7 μM substrate DNA in the absence of
calcium produces no observable effect on the stability reported
by Cys-36 (Fig. 5C). Following addition of 1 mM calcium, the
substrate DNA is rapidly degraded (inset Fig. 5C), and the
0.9 ? 0.1 kcalmol−1, corresponding to an apparent binding affi-
nity of 1.4 ? 0.4 μM, which is nearly identical to the affinity of the
inhibitor pdTp in the presence of 1 mM Ca2þand therefore is
presumably due to the effect of product binding.
The increasing trend of biological research towards more quan-
titative descriptions and models has generated an urgent demand
for simple and accessible techniques that can provide thermody-
namic data on such fundamental properties as protein stability
and ligand binding. We have demonstrated that quantitative cy-
steine reactivity can be used to determine protein stability using
only picomole quantities of material (nanograms for an average-
sized protein), readily accessible gel electrophoresis equipment,
and freely available gel analysis software. Furthermore, QCR as-
sesses stability at low protein concentrations, thereby minimizing
aggregation, a common problem in stability measurements made
by less sensitive methods. QCR exploits the fundamental rela-
tionship between protein flexibility and stability by monitoring
the differential reactivity of internal chemical groups in the native
and unfolded state, first pioneered by hydrogen exchange (HX)
as the experimental observation (5). Unlike HX, QCR observa-
tions are always obtained within the GUWO (∼10–15°C of Tm),
where global unfolding events dominate and where the reported
energetics correspond to global unfolding free energies.
QCR can be used to investigate many aspects of biological
function that are linked to protein stability. For instance, pro-
tein-ligand interactions can be readily identified and quantified
through the fundamental thermodynamic linkage relationships
between ligand binding and protein stability. This analysis can
be extended to infer enzymatic activity by monitoring changes
in stability in the presence of substrate, product (produced in
the course of the reaction), or inhibitors. Such observations by
themselves do not provide direct evidence of catalytic activity
but can be invaluable for establishing substrate specificity and in-
hibitor identity, even in the absence of reaction-specific kinetic
assays. The ability to obtain thermodynamic measurements with
small amounts of material and simple instrumentation enables
nique for protein characterization, including protein engineering
experiments and functional genomic studies that require the
thermodynamic characterization of a large number of variants.
Cell-Free Expression and Purification of Proteins Encoded by Synthetic Genes.
The wild-type proteins and cysteine variants were produced by cell-free in
vitro transcription and translation (TnT) using an E. coli extract from Bl21 Star
(DE3) (Invitrogen; C6010-03) (34) programmed with a synthetic linear DNA
fragment that was constructed using automated, PCR-mediated gene assem-
bly (35). The synthetic gene sequences (see SI Fabricated Synthetic Gene
Sequences (ORFs)) comprise a 5′ T7-promoter, a 5′ ribosome binding site,
and a 3′ T7-terminator flanking an open reading frame whose DNA
sequence was optimized for protein expression using a computational algo-
rithm that manipulates mRNA structure (Allert, Cox, and Hellinga, in prepara-
tion). All proteins contain a C-terminal FLAG-affinity tag (GGSDYKDDDDK)
(36) for purification. The version ofecRBP used in this study hasthe additional
mutation T3W. Approximately 2 μg of DNA was added to 200 μL TnT extract
and incubated at 30°C for 2 h. Proteins were purified using FLAG-affinity
beads (Sigma; F2426): beads were preblocked for 2 h (Starting Block; Thermo
Scientific; 37543) and washed with Tris buffered saline (TBS) (25 mM Tris,
150 mM NaCl, pH 7.4). Next, 100 μL of TnT extract was combined with
1 mL Flag beads (A600¼ 0.25), incubated at 4°C (15 min with end-over-
end mixing), washed twice with 1 mL of TBS, and eluted with 3x-FLAG
peptide (Sigma; F4799) buffer (25 mM MOPS, 100 mM KCl, 150 μM 3x-FLAG
peptide, and pH 7.6). Purified proteins were used directly in QCR ex-
The QCR Experiment. The rate of labeling of internal cysteine residues was
∼3–5 picomoles of protein) with IAM-biotin (EZ-link Iodoacetyl-PEG2-Biotin;
Pierce; 21334) in excess (1 mM unless otherwise stated) at constant tempera-
ture (25 mM MOPS, 100 mM KCl and pH 7.6). Five μL aliquots were removed
at fixed time intervals and quenched by addition of 2 μL 2 M β-mercaptoetha-
nol (Sigma; M6250). Following addition of 5 μL LDS-buffer (Invitrogen;
NP0007) and heating for 2 min at 85°C, (un)labeled protein species were re-
solved by SDS-PAGE (Novex 4–12% Bis-Tris Gels; Invitrogen; NP0321). Ob-
served gel shifts of the biotinylated species are caused either by slight
differences in conformation between the (un)labeled species (Fig. 1B) or,
more typically, by addition of 4 μL of 40 mg/mL streptavidin (Pierce;
21125) after heating (Fig. 1A). Following staining with GelCode Blue Stain
reagent (Thermo Scientific; 24592), gel images were digitized and band in-
tensitiesquantified by densitometry [ImageJ (37)] and fit to a single exponen-
tial to derive klabel(Tables S1 and S2). When designing QCR experiments, it is
important to consider the degradation of the iodoacetyl moiety of IAM-
biotin, which is dependent on time, exposure to light, and temperature.
At temperatures less than ∼65°C, we have observed this effect to be negli-
gible over the time-course of ∼2 hours. At higher temperatures the degrada-
tion of IAM-biotin must be taken into account, primarily by limiting the
labeling reaction to less than ∼90 min.
It is important to note that Eq. 4 applies only if the labeling conditions are
fully in the EX2 limit. As labeling reagent concentration increases, kintin-
creases concomitantly and eventually kintwill become equal to or greater
than kclose. Under these conditions, the observed labeling rate is determined
solely by kopenand klabeland is no longer a measure of stability. The reagent
concentration and environmental conditions (i.e. pH and temperature) at
which EX2 conditions no longer apply varies according to the (un)folding
kinetics of the protein. A simple test of reaction mechanism is to change re-
agent concentration and remeasure the kinetics: EX2 conditions are satisfied
if the change in the observed labeling rate is proportional to the change in
reagent concentration (Fig. S3).
∼0.1 μM proteinsample (i.e.,
Determining Intrinsic Reaction Rates of Unprotected Cysteines. The iodoacetyl
moiety of IAM-biotin reacts primarily with free thiolate to form a stable
thioether bond. Because thiolate is the predominant reactive species, the re-
action rate is dependent on the pKaof cysteine (∼8.6) relative to solution pH
(which is set to 7.6). It is important to note that the kintvalues reported here
are only valid for pH 7.6. kintwas determined for an unprotected cysteine by
reacting IAM-biotin with L-glutathione (GSH) (Sigma; G4251) and monitoring
the absorbance of the liberated iodide ion (ε226¼ 12;600 M−1cm−1at
226 nm) as a function of time. Second-order rate constants for the reaction
of IAM-biotin with GSH were measured under pseudo-first-order conditions
at 25°C, 35°C, 45°C, and 55 °C (80 μM IAM-biotin, 800 μM GSH, 25 mM MOPS,
100 mM KCl, and pH 7.6) and analyzed in terms of the Arrhenius equation
(Fig. S2). The slope (−Ea∕R) and preexponential factor (lnA) were found to be
−8.2 × 103? 0.1 × 103and 27.1 ? 0.5 s−1, respectively, enabling kintto be cal-
culated at any temperature. kintvalues derived from unfolded proteins were
in direct agreement with the GSH-derived kintvalues.
ACKNOWLEDGMENTS. This work was supported by the NIH Director’s Pioneer
Award (5DPI OD000122) and the Homeland Security Advanced Research
Projects Agency (HSH ODC-08-C-00099).
1. Wyman J (1964) Linked functions and reciprocal effects in hemoglobin: A second
look. Adv Protein Chem 19:223–286.
2. Schellman JA (1975) Macromolecular Binding. Biopolymers 14(5):999–1018.
3. Wyman J, Gill SJ (1990) Binding and Linkage (University Science Books, Mill Valley, CA)
4. Huyghues-Despointes B, Pace CN, Englander SW, Scholtz JM (2001) Protein Structure,
Stability, and Folding (Humana Press Inc, Totowa, NJ).
5. Hvidt A, Neilson SO (1966) Hydrogen exchange in proteins. Adv Protein Chem
6. Bai YW, Milne JS, Mayne L, Englander SW (1994) Protein stability parameters
measured by hydrogen-exchange. Proteins 20(1):4–14.
7. SwintKruse L, Robertson AD (1996) Temperature and pH dependences of hydrogen
exchange and global stability for ovomucoid third domain. Biochemistry-US
www.pnas.org/cgi/doi/10.1073/pnas.0910421107Isom et al.
8. Huyghues-Despointes BMP, Scholtz JM, Pace CN (1999) Protein conformational stabi- Download full-text
lities can be determined from hydrogen exchange rates. Nat Struct Biol 6(10):910–912.
9. Ghaemmaghami S, Fitzgerald MC, Oas TG (2000) A quantitative, high-throughput
screen for protein stability. Proc Natl Adac Sci USA 97(15):8296–8301.
10. Ha JH, Loh SN (1998) Changes in side chain packing during apomyoglobin folding
characterized by pulsed thiol-disulfide exchange. Nat Struct Biol 5(8):730–737.
11. FengZY, ButlerMC, AlamSL,Loh SN(2001) On thenatureof conformationalopenings:
Native and unfolded-state hydrogen and thiol-disulfide exchange studies of ferric
aquomyoglobin. J Mol Biol 314(1):153–166.
12. Sridevi K, Udgaonkar JB (2002) Unfolding rates of barstar determined in native and
low denaturant conditions indicate the presence of intermediates. Biochemistry-US
13. Jha SK, Udgaonkar JB (2007) Exploring the cooperativity of the fast folding reaction
of a small protein using pulsed thiol labeling and mass spectrometry. J Biol Chem
14. Silverman JA, Harbury PB (2002) The equilibrium unfolding pathway of a ðβ∕αÞ8barrel.
J Mol Biol 324(5):1031–1040.
15. Park C, Marqusee S (2004) Probing the high energy states in proteins by proteolysis.
J Mol Biol 343:1467–1476.
16. Bai YW, Englander JJ, Mayne L, Milne JS, Englander SW (1995) Methods in Enzymol-
ogy, ed Johnson ML (Academic, New York), Vol 259, pp 344–356.
17. Milne JS, Xu YJ, Mayne LC, Englander SW (1999) Experimental study of the protein
folding landscape: Unfolding reactions in cytochrome c. J Mol Biol 290(3):811–822.
18. Privalov P (1979) Stability of proteins. Adv Protein Chem 33:167–241.
19. Chaires JB (1997) Possible origin of differences between van’t Hoff and calorimetric
enthalpy estimates. Biophys Chem 64(1-3):15–23.
20. Prabhu NV, Sharp KA (2005) Heat capacity in proteins. Ann Rev Phys Chem 56:521–548.
21. Gomez J, Hilser VJ, Xie D, Freire E (1995) The heat-capacity of proteins. Proteins
22. Razvi A, Scholtz JM (2006) Lessons in stability from thermophilic proteins. Protein Sci
23. Rees D, Robertson AD (2001) Some thermodynamic implications for the thermostabil-
ity of proteins. Protein Sci 10:1187–1194.
24. Talla-Singh D, Stites WE (2008) Refinement of noncalorimetric determination of the
change in heat capacity, ΔCp, of protein unfolding and validation across a wide
temperature range. Proteins 71(4):1607–1616.
25. Prajapati RS, Indu S, Varadarajan R (2007) Identification and thermodynamic charac-
terization of molten globule states of periplasmic binding proteins. Biochemistry-US
26. Green SM, Meeker AK, Shortle D (1992) Contributions of the polar, uncharged amino-
acids to the stability of Staphylococcal nuclease—evidence for mutational effects on
the free-energy of the denatured state. Biochemistry-US 31(25):5717–5728.
27. Prajapati RS, et al. (2007) Thermodynamic effects of proline introduction on protein
stability. Proteins 66(2):480–491.
28. Kim JS, Kim H (1996) Stability and folding of a mutant ribose-binding protein of
Escherichia coli. J Protein Chem 15(8):731–736.
29. Bava KA, Gromiha MM, Uedaira H, Kitajima K, Sarai A (2004) ProTherm, version 4.0:
Thermodynamic database for proteins and mutants. Nucleic Acids Res 32:120D–121D.
30. Waldron TT, Murphy KP (2003) Stabilization of proteins by ligand binding: Application
to drug screening and determination of unfolding energetics. Biochemistry-US
31. Serpersu EH, Shortle DR, Mildvan AS (1986) Kinetic and magnetic-resonance studies of
effects of genetic substitution on Ca2þ-liganding amino-acid in Staphylococcal
nuclease. Biochemistry-US 25(1):68–77.
32. Pace CN, McGrath T (1980) Substrate stabilization of lysozyme to thermal and guani-
dine-hydrochloride denaturation. J Biol Chem 255(9):3862–3865.
33. Cuatrecasas P, Fuchs S, Anfinsen CB (1967) Catalytic properties and specificty of
extracellular nuclease of Staphylococcus aureus. J Biol Chem 242(7):1541–1547.
34. Jewett M, Swartz J (2004) Mimicking the Escherichia coli cytoplasmic environment
activates long-lived and efficient cell-free protein synthesis. Biotechnol Bioeng
35. Cox JC, Lape J, Sayed MA, Hellinga HW (2007) Protein fabrication automation.
Protein Sci 16(3):379–390.
36. Hopp T, et al. (1988) A short polypeptide marker sequence useful for recombinant
protein identification and purification. Nat Biotechnol 6(10):1204–1210.
37. Rasband W (1997–2009) ImageJ (U.S. National Institutes of Health, Bethesda,
38. Hale S, Poole L, Gerlt J (1993) Mechanism of the reaction catalyzed by Staphylococcal
nuclease: Identification of the rate-limiting step. Biochem J 32(29):7479–7487.
Isom et al. PNAS
March 16, 2010